chore: convert generators (generate_patt + nft_marker_gen) to arlog_*! (PR 3/4 for #90)

[maxwelljhuang]

Part 3 of 4 for #90 . Converts the two generator CLI examples to the project's arlog_*! macros. PR 3 is larger than PR 1 or PR 2 (84 print sites combined) and exercises arlog_e! and arlog_w! in examples/ for the first time, plus introduces three "do-not-convert" categories: CLI help banners, interactive prompts, and empty spacers. Calling out the contestable judgments up front so review can focus there.

Scope

• crates/core/examples/generate_patt.rs — 716 lines, 34 print sites: 16 help-banner eprintln! kept, 16 → arlog_i!, 1 → arlog_w!, 1 → arlog_e!. Logger init + imports added.
• crates/core/examples/nft_marker_gen.rs — 399 lines, 50 print sites: 6 y/N prompt sites kept, 10 empty spacers deleted, 25 → arlog_i!, 7 → arlog_e!, 2 → arlog_w!. Logger init was already present from a prior change; imports added.
• crates/core/Cargo.toml — new [[example]] block for generate_patt with required-features = ["log-helpers"]. nft_marker_gen's block already exists.

Categorical decisions

CLI help banner stays as eprintln! (16 sites in generate_patt.rs print_help_and_exit). Routing --help output through arlog_i! would prefix every line with [info] and apply level filtering — wrong UX for help text, which is human UI not log events.

Interactive y/N prompt stays as eprintln!/eprint! (6 sites in nft_marker_gen.rs). The prompt + read_line block needs synchronous stderr output that a log facade can't guarantee.

Empty println!(); spacer lines deleted rather than converted (10 sites in nft_marker_gen.rs). An empty arlog_i!() renders as a bare [info] with no body, which is noise. The spacing was a stdout-style artifact from the original C markerCreator. Happy to preserve them in some other form (e.g. arlog_i!("") to maintain visual structure) if you'd prefer — let me know.

if cfg.debug / if cfg.verbose wrappers in generate_patt.rs preserved, inner calls converted to arlog_i! with "Debug:"/"Verbose:" prefixes intact. Two cleaner alternatives exist:

• (b) wire cfg.debug/verbose to set_ar_log_level() at startup, drop wrappers, use arlog_d!. Behavior-preserving but introduces global state coupling and would unmute any arlog_d! in linked library code.
• (c) drop the CLI flags entirely, use arlog_d!, document RUST_LOG=debug. Cleanest semantics, but a user-visible CLI surface change.

Defaulting to (a) for minimum-change. Happy to flip to (b) or (c) — both are reasonable, and (c) might be the right long-term answer; let me know your preference and I'll push a fixup or open a follow-up.

Sub-classification: lines 328, 341, 347 in nft_marker_gen.rs are three per-scale outcomes in the FREAK feature loop, classified differently. 328 (arlog_i!) is success, 341 (arlog_i!) is an expected skip for small scales, 347 (arlog_w!) is an unexpected per-scale failure where the loop continues. Following the existing author's signal (println vs eprintln) for which is which.

small adjustment beyond the plan

use webarkitlib_rs::arlog_w; in nft_marker_gen.rs is gated on #[cfg(feature = "ffi-backend")] to silence an unused-import warning on the no-ffi build. Both arlog_w! call sites are already inside that cfg block; the import follows them.

Verification

• cargo fmt --all -- --check ✓
• cargo build --example generate_patt --features log-helpers ✓
• cargo build --example nft_marker_gen --features log-helpers ✓
• cargo clippy --features log-helpers — no new lints in either touched file
• cargo test ✓
• cargo run --example generate_patt --features log-helpers -- --help ✓ — confirmed plain text output, no [info] prefixes (Q1 verified)
• cargo run --example generate_patt --features log-helpers ✓ — confirmed [info] arlog output on the converted paths
• cargo run --example nft_marker_gen --features log-helpers (with -y to skip the y/N prompt) ✓
• grep for residual print calls — only the 16 help-banner sites and 6 y/N prompt sites remain, as planned

Skipped per pre-existing issues already documented on the issue thread: cargo build --all-features (macOS stdc++ link issue) and cargo clippy --all-targets -- -D warnings (lib-side doctest lint in arlog.rs:299).

Followups

• PR 4: simple_nft.rs (depends on ffi-backend feature, kept separate so reviewers without an FFI setup could sign off on PRs 1–3 independently).

[kalwalt]

Thank you for this PR! 🙌🔝 i will review It this afternoon.

[kalwalt]

You can use the arlog_rel here and in the lines after this code printing, it will not print the [info] pre text or other type of header.

[kalwalt]

As i said for the generate_patt.rs you can use arlog_rel here. Don not appyly to this line eprint!("Continue anyway? [y/N] "); i'm not sure that arlog_rel can works.

[kalwalt]

I left some comments in the code, if you can apply the changes requested. After that i will merge it.